随机计时离散事件系统的软控制任务研究

本文通过引入马氏决策过程中的迭代算法,研究了计时离散事件系统的随机优化监控综合问题。为了对不确定的人造系统实施监控,在考虑事件的操作时间的基础上,利用带有发生事件概率分布函数的随机计时离散事件系统模型对系统建模。为了对这类随机系统实施监控,在传统方法中,采用控制任务的最大可控子语言设计控制器,不能体现系统模型的随机特性。本文提出利用软控制任务代替原控制任务的方法,使其超出原控制任务的概率在给定的容许度约束范围内。首先,通过在计时离散事件系统中定义计时事件的发生概率映射和发生费用函数,利用离散事件系统的逻辑特性,构造事件发生序列的期望费用函数,进而确立马氏决策过程的最优方程,建立软控制任务与期望费用函数之间的关系。然后,通过计算事件发生序列的费用值,提出利用有限费用值可以用来确定软控制任务,进而基于逻辑监控方法,确定最优监控器。最后,利用计算有限费用值的迭代过程,提出迭代算法,并给出了计算实例。     

组件多工作模式下的离散事件系统监督控制方法研究

针对组件多工作模式下的离散事件系统,提出一种监督控制方法.利用Ramadge-Wonham监督控制架构,计算出组件各工作模式下对应的监督控制器.然后,利用所提的merge算法将组件各工作模式下对应的监督控制器合并,以生成融合监督控制器.再利用所提的事件选择函数得到在融合监督控制器各状态允许发生的事件,从而保证系统在融合监督控制器作用下的可控且非阻塞运行.最后,通过一个带反馈功能的制造系统演示本文所提方法的有效性,该系统中的检测单元将根据工件未通过检测的次数选择工作模式.相比于运用扩展有限状态机进行变量抽象的方法,本文所提方法更加直观简便,并能适应多个工件同时处于系统中时的情形.     

DoS攻击下网络化控制系统基于观测器的记忆型事件触发预测控制

本文研究了DoS攻击下网络化控制系统记忆型事件触发预测补偿控制问题. 首先, 由于网络带宽资源有限 和系统状态不完全可观测性, 引入了记忆型事件触发函数, 为观测器提供离散事件触发传输方案. 然后, 分析了网络 传输通道上发生的DoS攻击. 结合上述记忆型事件触发方案, 在控制节点设计一类新颖的预测控制算法, 节省网络 带宽资源并主动补偿DoS攻击. 同时, 建立了基于观测器的记忆型事件触发预测控制的闭环系统, 并且分析稳定性. 通过线性矩阵不等式(LMI)和Lyapunov稳定性理论, 建立了控制器、观测器和记忆型事件触发矩阵的联合设计方案, 并验证了该方案的可行性. 仿真结果表明, 该方案结合记忆型事件触发机制可以有效补偿DoS攻击, 节约网络带宽 资源.     

基于状态树结构的离散事件系统模块化监督控制

在状态树结构(State tree structures, STS)的基础上提出了离散事件系统的模块化监督控制方法. 该方法中, 系统模型是状态树结构模型, 控制指标以谓词形式给出. 把控制指标分解为几个子控制指标的合取形式, 对每个子控制指标分别设计非阻塞最优监督控制器, 并保证闭环系统行为满足控制指标要求且是非阻塞最优的. 本文研究了模块化监督控制器存在性条件, 给出了模块化监督控制器的综合过程; 当闭环系统阻塞时, 引入协调器来解决闭环系统阻塞问题并且给出了协调器的设计方法.     

双事件触发下非匹配干扰系统积分滑模抗干扰跟踪控制

对于带有不匹配干扰的动态系统,设计具有高资源利用率的抗干扰控制算法具有重要意义.本文在双事件触发机制框架下,研究了非匹配干扰系统的积分滑模抗干扰控制以及动态性能分析问题.首先,基于增广模型构建干扰观测器,实现对未知不匹配干扰的动态估计.为了降低数据冗余并保证传输的同步性,以反馈状态及干扰估计的单触发条件为基础,本文构建了“先到同触发”的双事件触发理论框架.在此框架下,设计积分滑模面和对应的双触发抗干扰控制器,保证被控系统的状态收敛到滑模面.基于Lyapunov稳定性分析方法,计算控制器和观测器增益,保证增广闭环系统具有良好的稳定性和动态跟踪性能.进一步分析了由触发引起的Zeno现象将不会发生.最后,基于典型的A4D模型进行仿真验证,仿真结果表明本文所提的方法具有良好的抗干扰性能.     

切换非线性系统的输出反馈周期事件触发控制

本文针对一类在任意切换信号作用下的切换非线性系统, 研究了其输出反馈周期事件触发控制问题. 所考 虑的非线性系统采用非严格反馈形式且含有未知时变控制系数. 在本文中, 仅利用采样时刻的系统输出. 为了估计 系统的不可量测的状态, 基于采样的系统输出构造了降维状态观测器. 为了减少通信资源的利用, 提出了一种新的 输出反馈周期事件触发策略, 该策略包含仅利用事件触发时刻的信息构造的输出反馈事件触发控制器以及仅在采 样时刻间歇性监测的离散事件触发机制. 通过选取可容许的采样周期及合适的公共Lyapunov函数, 证明了闭环系统 在任意切换下全局渐近稳定. 最后, 通过将本文中所给出的控制方案应用到数值算例中验证了其有效性.     

基于观测器的周期拒绝服务攻击网络化系统动态事件触发控制

针对具有周期拒绝服务(DoS)攻击的网络化系统,设计一种基于观测器的具有动态事件触发策略的控制器.首先,通过DoS攻击对网络化系统的影响建立了DoS攻击模型,采用切换系统的方法,将具有DoS攻击的网络化系统分为DoS攻击活跃子系统和DoS攻击休眠子系统.对不可测的系统状态设计状态观测器,通过在静态事件触发中引入一个内部...     

连续计时离散事件系统监控及其可观性

本文提出了一种带有连续时间变量的离散事件系统(称为计时离散事件系统)结构模型.通过讨论计时语言的性质,如封闭性、可控性以及可观性,研究了计时离散事件系统的监控综合问题,并基于这些性质,分别提出了计时离散事件系统在完全可观与部分可观条件下监控器存在的充要条件.     

一类线性切换系统的事件触发输出调节问题

通过设计事件触发状态反馈控制器,研究一类线性切换系统的输出调节问题.相比于时间触发控制,事件触发控制可显著地降低控制任务执行的次数.由于事件触发时刻与切换时刻的相互混合,导致切换系统控制器的设计十分困难.本文通过给出事件触发机制、状态反馈控制器和满足平均驻留时间的切换信号联合设计的方案,得到系统输出调节问题可解的充分条件.非共同坐标变换突破原有各子系统的调节器方程组具有共同解的限制.此外,为了避免Zeno现象,给出相邻事件触发间隔时间存在一个正下界.最后,应用切换RLC电路系统进行仿真验证结论的有效性.     

不确定非线性系统自适应动态事件触发输出反馈镇定

本文研究了一类不确定非线性系统的动态事件触发输出反馈镇定问题. 显著不同的是系统具有依赖于不可测状态的增长且增长率为输出的未知多项式. 尽管已有一些连续自适应控制器, 但需要巧妙融合非线性状态观测器、系统未知性的动态补偿以及非线性的抵御, 因此这些控制器具有一定的脆弱性, 不能平凡地拓展到不连续情形 (采样误差导致). 为此, 首先通过引入动态高增益和基于高增益的观测器来分别抵御未知增长率和重构系统不可测状态. 进而, 意识到静态事件触发机制的无效性, 通过引入动态事件触发机制, 成功设计出了事件触发输出反馈控制器, 确保了系统状态的全局有界性和收敛性. 数值仿真验证了所设计控制器的有效性.     

Supervisory control of discrete event systems with communication delays and partial observations

This paper presents necessary and sufficient conditions for the existence of a nonblocking supervisor that achieves a given language specification for a discrete event system (DES) with communication delays and partial observations. In many practical situations, some uncontrollable events can subsequently occur before a proper control action is applied to the DES due to delays in sensing, communicating, and actuating. Moreover, some of the uncontrollable events may be unobservable. To achieve a given language specification in such situations, this paper presents a language property called delay observability which assures no confliction in making a decision for legal controllable events under partial observation and delay communication.     

基于Petri网结构分析的监控器综合

在基于Petri网建模的离散事件系统中, 提出利用局部关联信息进行约束转换, 并实现Petri网结构监控器综合的方法. 对以Parikh矢量约束形式给出的控制规范, 不可控不可观变迁会导致约束成为非法约束, 分析了不可控变迁的前向关联结构和不可观变迁的后向关联结构, 利用局部关联变迁实现对不可控和不可观变迁的间接控制, 从而将非法矢量约束转换为合法约束, 并保证初始控制规范的实现. 与基于矩阵的监控器综合方法相比, 本文的方法只需利用局部信息, 最后通过实例对该方法进行了说明.     

Optimal non‐blocking decentralized supervisory control with unobservable controllable events

Paranormality is an observation property of a language, in which the occurrence of unobservable events never exits the closure of the language. In this paper, a synthesis method is proposed to construct a paranormal supervisor. We propose a method to construct a controllable language such that the occurrence of unobservable events does not exit the closure of the controllable language. Moreover, a new observation property, that is, Quasi Output Control Consistency (QOCC) is defined to construct the optimal (least restrictive) non‐blocking decentralized supervisory control in the presence of unobservable controllable events. Using QOCC and natural observer properties, we propose a method to construct a normal supervisor such that an arbitrary pair of lookalike strings are initiated and terminated with identical observable and uncontrollable events. It is assumed that one of these strings has unobservable controllable events. An OCC property is defined in the literature as a special case of QOCC property, where none of the lookalike strings has unobservable controllable events.     

On optimal control of a class of partially observed discrete event systems

We are interested in a new class of optimal control problems for discrete event systems. We adopt the formalism of supervisory control theory (Proc. IEEE 77(1) (1989) 81) and model the system as a finite state machine (FSM). Our control problem is characterized by the presence of uncontrollable as well as unobservable events, the notion of occurrence and control costs for events and a worst-case objective function. We first derive an observer for the partially unobservable FSM, which allows us to construct an approximation of the unobservable trajectory costs. We then define the performance measure on this observer rather than on the original FSM itself. We then use the algorithm presented in Sengupta and Lafortune (SIAM J. Control Optim. 36(2) (1998)) to synthesize an optimal submachine of the C-observer. This submachine leads to the desired supervisor for the system.     

Robust and nonblocking supervisor for discrete-event systems withmodel uncertainty under partial observation

Addresses a robust and nonblocking supervisory control problem based on a framework of discrete-event systems with model uncertainty under partial observation. The uncertainty introduced in the paper is associated with internal and unobservable events occurring in systems. The paper presents a systematic method for modeling uncertainty described as a Δ-transition representing the internal and unobservable events. In particular, at a state with the assigned positive integer value p a system is assumed to experience at most a p-step state transition by the internal and unobservable event. Under the framework for model uncertainty, the paper presents necessary and sufficient conditions for the existence of a robust supervisor with the nonblocking property     

Synthesis of supervisory controllers for hybrid systems based onapproximating automata

The paper concerns the synthesis of supervisory controllers for a class of continuous-time hybrid systems with discrete-valued input signals that select differential inclusions for continuous-valued state trajectories and event-valued output signals generated by threshold crossings in the continuous state space, the supervisor is allowed to switch the input signal value when threshold events are observed. The objective is to synthesize a nonblocking supervisor such that the set of possible sequences of control and threshold event pairs for the closed-loop system lies between given upper and lower bounds in the sense of set containment. We show how this problem can be converted into a supervisor synthesis problem for a standard controlled discrete-event system (DES). A finite representation may not exist for the exact DES model of the hybrid system, however. To circumvent this difficulty, we present an algorithm for constructing finite-state Muller automata that accept outer approximations to the exact controlled threshold-event language, and we demonstrate that supervisors that solve the synthesis problem for the approximating automata achieve the control specifications when applied to the original hybrid system     

Supervisor direct synthesis method for a structured discrete dynamical system

We present an implementation method for a supervisor of a discrete event system (DES) structured in a special way. Such a DES has the following characteristic features: the language generator is represented by a set of finite automata, the language model is extended with a new type of events called “expected,” and the specification is defined as a sequence of control commands. The primary advantage of our method is that it uses structural knowledge of the event stream in order to construct a supervisor. Thus, we achieve linear dependence of the supervisor’s size in the input data.     

Control of Non-Deterministic Systems With μ-Calculus Specifications Using Quotienting

The supervisory control problem for discrete event system(DES) under control involves identifying the supervisor, if one exists, which, when synchronously composed with the DES,results in a system that conforms to the control specification. In this context, we consider a non-deterministic DES under complete observation and control specification expressed in action-based propositional μ-calculus. The key to our solution is the process of quotienting the control specification against the plan resulting in a new μ-calculus formula such that a model for the formula is the supervisor. Thus the task of control synthesis is reduced a problem of μ-calculus satisfiability. In contrast to the existing μ-calculus quotienting-based techniques that are developed in deterministic setting, our quotienting rules can handle nondeterminism in the plant models. Another distinguishing feature of our technique is that while existing techniques use a separate μ-calculus formula to describe the controllability constraint(that uncontrollable events of plants are never disabled by a supervisor), we absorb this constraint as part of quotienting which allows us to directly capture more general state-dependent controllability constraints. Finally, we develop a tableau-based technique for verifying satisfiability of quotiented formula and model generation. The runtime for the technique is exponential in terms of the size of the plan and the control specification. A better complexity result that is polynomial to plant size and exponential to specification size is obtained when the controllability property is state-independent. A prototype implementation in a tabled logic programming language as well as some experimental results are presented.     

Safe Supervisory Control Under Observability Failure

This paper discusses about supervisory control under possible loss of observability. The final goal of this work is to design a safe (avoiding disaster states) supervisory control loop structure taking into consideration the uncertainty that stems from the fact that some event(s) may turn into unobservable at some point along the normal system operation. This kind of failure may correspond to the breakdown of some plant sensor. The supervisor finally obtained is in general: a) more permissive than the supervisor obtained assuming those events as unobservable from the beginning, since at some points it shall be better informed, and b) more restrictive than the supervisor obtained assuming that those events shall never fail, since it will have to prevent the system from following some undesirable trajectories that the system could take in presence of an observability failure. This paper presents results to obtain a safe controller that avoids disaster states in presence of the described uncertainty, and also ensures that the system behaviour will not run out from its specifications in absence of failure.     

Progressive events in supervisory control and compositional verification

This paper investigates some limitations of the nonblocking property when used for supervisor synthesis in discrete event systems. It is shown that there are cases where synthesis with the nonblocking property gives undesired results. To address such cases, the paper introduces progressive events as a means to specify more precisely how a synthesised supervisor should complete its tasks. The nonblocking property is modified to take progressive events into account, and appropriate methods for verification and synthesis are proposed. Experiments show that progressive events can be used in the analysis of industrial-scale systems, and can expose issues that remain undetected by standard nonblocking verification.